See also: IRC log
http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0031.html
<PaulC> http://www.w3.org/Bugs/Public/show_bug.cgi?id=3672
<Yakov> http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0073.html
paul: is there anyone that needs this text explained? people ready to vote/adopt?. consensus to adopt
RESOLUTION: 3672 adopted http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0073.html
<asir> related editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/33
http://www.w3.org/Bugs/Public/show_bug.cgi?id=3621
http://lists.w3.org/Archives/Public/public-ws-policy/2006Aug/0145.html
bijan: if there's really no interest in WG suggests to drop it. found some of the text ambiguous
<dorchard> I think formal semantics might be useful IFF it helped interoperability. Then it becomes a trade-off between working on formal semantics to help interop/clarity, or some other means like test cases.
paul: work has some value, but it has to be work that whole WG to take on. typically it's subset of WG. encourage participants to raise issues related to formal semantics
ashok: worth spending time on bijan's work on policy. takes policy statements and translates them to rdf. can generate policy models
<bijan> technically to owl
dan: that's cool. has there been any issues?
ashok: bijan has demo
daveo: seems a tradeoff in mechanism between clarity and interop. in general semantic work can help, but worries about tradeoff
dan: if there's a mapping of policy to rdf, then doesn't rdf define formal semantics to policy?. if that work is done, what remains to be done?
paul: he hasn't found time to report that work. not seeing people clamoring that WG needs to add semantic work. asks bijan if he wants to take a formal AI
bijan: i'll do it on my own time
RESOLUTION: Close 3621 with no changes
http://lists.w3.org/Archives/Public/public-ws-policy/2006Aug/0181.html
bijan: ability to express relations between assertions is important
<dorchard> glen, is this kind of like the re-use part of Features in F&P?
daveo: wondering if this is similar to reuse part of features and properties
glen: yes, definitely. one of the things that was intended. not the same thing, but reflects similar ideas
asir: thought it was related to 3621, formal semantics, rdf. is that fair?
bijan: i think you misunderstand the issues. here we are talking about relationships between assertions, not policies in general
more about specific functionalities
paul: is there support for doing this work?
ashok: if you have an assertion with parameters, they all have the same qnames. is that what bijan wants?
paul: example about RM. relationship of those things is then hidden inside parametric nature. not seeing support for this work
RESOLUTION: Issue 3622 Closed (NotFixed)
<dorchard> For the record, BEA was very slightly interested in pursuing 3622(re-use/generality) as a way of providing additional re-use mechanism, and this is an candidate for V.Next
paul: related to 3694. would rather process it then
http://www.w3.org/Bugs/Public/show_bug.cgi?id=3602
<PaulC> http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0094.html
paul: ashok took an AI for a revised proposal
ashok: we had a lot of discussion on this. turns out thanks to maryann that we figured out the actual text was exactly right. actual text conveys semantics. only problem is that one sentence difficult to understand. spelt it out with an example
monica has suggestion
http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0099.html
ashok: monica suggests it's too early in the document. with recast, it's not obvious what intent is, so example is required
monica: we have material later in document to address "optional". guideline to give guidance on how to use
maryann: appreciate the spirit, but change doesn't help clarify difference. forward reference in document is good idea. also guideline document to help, and include additional references
paul: what about a forward reference to the example under "optional". likes example, example helps understanding. that has worked in other WG's
asir: there is an example in 4.3.1
paul: could we add at end of 4.3.1?
prasad: example illustrates one scenario where assertion does not occur should be prohibited. we should give forward references to concept instead of moving example
fabian: would like to discuss original issue (not example location)
paul summarizes what occured so far on this issue
fabian: maybe just me not understanding second sentence. example - RM optional. after normalizing, two different policies that apply to same endpoint, one of which prohibits RM
paul: consensus is that is correct
dan: that's not what fabian is saying. asserts that this is causing confusion because impression is that client is prohibited
monica: still thinks this is for guidance document
vladb: what about introducing a new assertion?
maryann: one solution is a new binding that handles the new assertion and supports the old. wondering if changing "client" to "entities"
dan: people are reading the spec and interpreting it in two ways. we need them to interpret it in one way
maryann: what they're trying to indicate with alternatives/optional is that one alternative doesn't. and since you know about the other assertion and you don't include it, then it's prohibited
dan: that's how we understand it, but we need everybody else to understand it that way too
paul: possible compromise is to add monica's text and example
<Yakov> +1 to Maryann
ashok: spec speaks of situation where i have to select one policy alternative. when you use the word "provider" it looks like policy applies to one direction. wsp framwork has no direction. we don't speak of provider, requester, etc
paul: that's agenda item 29a
asir: supports monica's text with the changes, and provide example and references
monica: +1
yakov: deep reservations about text - talks about provider, and "applies" policy
<asir> changes to Monica's text are drop 'but'
yakov: not clear what that means about provider applying policy
<asir> second change to Monica's text is s/provider/provider of the service/
yakov: reluctant support, and changes in 3.2 section later
prasad: wanted to address ashok's point in case of optional assertion. it's up to the client to use which alternative since provider has no preference. then provider of service decides which alternative was selected
whenry: couldn't you just use monica's text and use "provider" and "consumer"?
maryann: don't know if there is consensus in the group, can we take another round?
monica: should go ahead and make decision then defer to 29a
dan: we want to allow people to try to send a message, even if they don't understand. word "prohibited" is at issue
paul: proposal on table is to add monica's text, example, and references. pushback is about using the "provider" (maryann, ashok, yakov). will changing "provider" fix things?
maryann: don't have words right now that can provide consensus. intent here is to clarify, and not sure if that clarifies
ashok: unhappy because it implies it works one way and does not work the other
<danroth> what about changing provider to policy subject?
ashok: policies work both ways
paul: problem is that piece of text. give dissenters an explicit AI to respond with an alternative by email. we should not constrain them to that phrase
ACTION: maryann, ashok, yakov to respond with alternative text [recorded in http://www.w3.org/2006/09/14-ws-policy-minutes.html#action01]
whenry: in terms of provider issue, would be useful to explain what they're afraid of (rationale). maybe issue doesn't exist
fred: has alternative. desire to record clearly current revised proposal, perhaps record concerns with it as issue
ashok: if we could get fabian to write email about why the old text wasn't right, that would help
toufic: to put in email issue about which policy alternative the "provider" selects to enforce
http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0099.html
<prasad> ashok: I request fabian to put his concern in a short email. His concern is not clear to me
<PaulC> http://www.w3.org/Bugs/Public/show_bug.cgi?id=3703
http://www.w3.org/Bugs/Public/show_bug.cgi?id=3703
<PaulC> http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0077.html
<PaulC> Change SHOULD to MUST
glen: two kinds of extensions - assertions, and extensions to the framework. in either case you should recognise the element
RESOLUTION: Accept as proposed in http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0077.html
<dorchard> glen, pls read the material I wrote for the primer on versioning
asir: this one is already done
<asir> related editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/34
paul: defn of "nested policy expression" in 2.4 of editors draft resolves this issue
RESOLUTION: Already done http://www.w3.org/2005/06/tracker/wspolicyeds/actions/34
http://www.w3.org/Bugs/Public/show_bug.cgi?id=3708
http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0076.html
fred: security considerations are informative. therefore should be "elevated" to the spec document
maryann: isn't that a topic for primer?
fred: most specs have a "security considerations" section. ok with primer if it wants to elaborate, but main point is that it should be in main document
paul: primer might want to point to section in main document. two part proposal: 1. move security considerations into the framwework doc
2. have primer point to that section
prasad: what about attachment doc?
fred: if there are issues unique to the framwork
ACTION: prasad to review next editors draft to see if there are any security considerations from framework doc that apply to attachment doc [recorded in http://www.w3.org/2006/09/14-ws-policy-minutes.html#action02]
<trackbot> Created ACTION-110 - Review next editors draft to see if there are any security considerations from framework doc that apply to attachment doc [on Prasad Yendluri - due 2006-09-21].
asir: clarification - it's going to be part of security considerations section
RESOLUTION: 3708 accept Frederick's proposal AND add pointer in primer
break
<asir> related editorial actions are
<asir> http://www.w3.org/2005/06/tracker/wspolicyeds/actions/35
<asir> http://www.w3.org/2005/06/tracker/wspolicyeds/actions/36
[Asir and Jonathan Marsh to present]
<asir> WSDL 20 is http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/wsdl20/wsdl20.html?content-type=text/html;%20charset=utf-8
<PaulC> http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0037.html
<PaulC> is the email proposal for this issue
Jonathan: section 9. XML syntax summary. port became endpoint. extensibility model in 2.0 is open content model. there is an order of element at top level. section 2.14 service component. section 2.3.1 interface faults. faults have been elevated to peers of operations. operations can refer to faults, and policies can be attached to faults
asir presents contribution: WS-Policy Attachment for WSDL 2.0
4 major pieces:
- enumerates attachment points
- 2. describes policy subjects
- defines an extension to the component model from wsdl 2.0
- explains how to calculate effective policy for each policy subject
ashok: clarification - not required to have policies in wsdl document?
asir: no, just an example
glen: do we make clear enough that the fact a policy exists inside a description doesn't apply to the entire description?
asir: no, there are only four policy subjects
glen: wondering if there is a term that can be introduced to mean "the merge of Policy or PolicyReference elements"?
paul: looking for a macro
asir: probably should use "element policy" as a macro (editorial action)
glen: section 4, eliminate second column and change table title to include {policy}
fred: having trouble understanding how to follow the diagram. inheritance?
asir: these are aggregations
paul: fred wants to know what arrows mean
fred: e.g. between binding and bindingFault
Jonathan: no significance to colour in arrows, just overlap issue
paul: might want to use the same colour for arrows (editorial action). do you think this is complete? any areas that you don't think are covered?
asir: thinks it's complete
ashok: when you're doing policy merging to compute effective policy, can you have errors if something goes wrong?
the merge is a very simple operation
asir: the merge is a very simple operation. it's a cross product, nothing more
ashok: suppose an assertion has parameters, and the identical assertion with different values for the parameter...
paul: is this a question about this proposal, or about the merge operation definition?
ashok: question is about merge definition
suppose an assertion has parameters, and the identical assertion with different values for the parameter...
<PaulC> Attachment defn of "merge":a merge consists of serializing each policy as a policy expression, replacing their wsp:Policy element with a wsp:All element, and placing each as children of a wrapper wsp:Policy element.
<PaulC> Attachment defn of "merge":
ashok: what happens?
asir: aggregate behaviour is delegated to domain. example SignedParts - specifics on how to interpret the merge behaviour
paul: section 3.2 in framework doc about aggregate behaviour
<PaulC> Mechanisms for determining the aggregate behavior indicated by the assertions (and their Post-Schema-Validation Infoset (PSVI) content, if any) are specific to the assertion type and are outside the scope of this document.
ashok: doc does not talk about using external attachment mechanism from wsdl 2.0
no, it doesn't do that. requirements that the doc satisfy are spelled out in section 1. Introduction
paul: question was asked yesterday in the context of proposal we have for 1.1. i said we can ask the same question about 2.0. when we adopted conformance text, was it only for framework?
asir: also attachment
paul: so same needs to be done for wsdl 2.0
ACTION: ashok to ask question/raise issue about what WG wants to do about attachments for WSDL 2.0 [recorded in http://www.w3.org/2006/09/14-ws-policy-minutes.html#action03]
<trackbot> Created ACTION-111 - Ask question/raise issue about what WG wants to do about attachments for WSDL 2.0 [on Ashok Malhotra - due 2006-09-21].
paul: do we believe that this proposal, modified by editorial changes, and addition of conformance text for wsdl 2.0 based on text from yesterday, resolves wsdl attachment issue?
RESOLUTION: 3694 accepted with proposal as amended
ACTION: paul to send an email to the chair of SAWSDL WG informing of the proposal adopted today [recorded in http://www.w3.org/2006/09/14-ws-policy-minutes.html#action04]
<trackbot> Created ACTION-112 - Send an email to the chair of SAWSDL WG informing of the proposal adopted today [on Paul Cotton - due 2006-09-21].
<asir> 3694 editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/37
agenda is at http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0041.html
http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0007.html
asir: 3705 deals with this issue
http://www.w3.org/Bugs/Public/show_bug.cgi?id=3705
when you select one policy alternative, does that apply to all messages between two participants?
asir: a message exchange may be described by a policy on policy subjects. also may be one for input, one for output, one for fault. multiple possibilities. all policies are chosen from the different policy subjects, which describes the exchange
ashok: reading of the spec is you select effective policy
glen: which portion of the spec?
maryann: intersection
intersection is an operation which takes in two policies and produces one
glen: the reality is that you have a hierarchical set of policies
ashok: maybe we should put words behind it?
asir: attachment spec, Section 4.1
glen: in WS model section there is mention of requester, provider, etc. for each message that is exchange, there is an agreed upon policy set
Jonathan: effective policy vs. relevant policy is confusing
paul: move on to Q2
glen: we said moot to the question - the notion of a single policy alternative applying to the whole interaction between client and server does not apply
paul: different selected policies at different levels
glen: authors of assertions should make clear the direction. guidelines should clarify
paul: Q3 is delayed - will deal with it on 3639
http://www.w3.org/Bugs/Public/show_bug.cgi?id=3619
<PaulC> http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0052.html
asir: talking about example in 3.4 (external attachment mechanisms). bob is asking for a normative description, not a fictitious example
paul: this is an example of normative material above. does the normative material address the issue, or are they asking us to add to the normative material
ashok: there is no normative material, just a statement that you may use it
paul: using EPR in appliesTo
this means to me that the scope of it is the EPR
paul: we can a sentence that clarifies that this is the scope
glen: don't believe that's sufficient
paul: we can try to answer that question as it applies to EPR. but that's an extensibility point. why do we do it for EPR?
ashok: because it's a particular usage
paul: our response maybe should be: we think you should, but in this case, here's an example
ashok: if we have done the work, why don't we put it in?
glen: because it might not apply to our spec. ok with doing the work, and deciding later where it goes
ACTION: glen to draft a response to the WS-A WG, and keep it on member list [recorded in http://www.w3.org/2006/09/14-ws-policy-minutes.html#action05]
<trackbot> Created ACTION-113 - Draft a response to the WS-A WG, and keep it on member list [on Glen Daniels - due 2006-09-21].
Fabian: I pointed out to Paul that I still have a question pending on issue 3619. I posted it to the mailing list yesterday. I will resend if nobody picks up on it in time. My email is archived at Re: Issue 3619 - input from WS-Addressing WG
RESOLUTION: Fixed deemed editorial, assigned to editors
<asir> related editorial action is http://www.w3.org/2005/06/tracker/wspolicyeds/actions/38
fred: text of proposal has two different topics. we can deal with this in two parts
<PaulC> http://www.w3.org/Bugs/Public/show_bug.cgi?id=3710
asir: a & b sound ok. I don't understand item 3 - what does it say?
paul: descendant is not a defined term. this is the best text we could capture
dan: this text is saying what we dont' do. this space is very large. if this text never existed, would anyone have a problem?
paul: combination with item a is what led us here
remember we're writing for an audience of people who will develop these policies, not just for us
maryann: it only says what it doesn't do about arbitrary processing, but it's specific about the rest. definitely support this
RESOLUTION: 3710 resolved by material in the bug report
http://www.w3.org/Bugs/Public/show_bug.cgi?id=3711
<FrederickHirsch> http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0100
<PaulC> Proposed text is: Similarly, by repeatedly distributing wsp:All over wsp:ExactlyOne,"
paul: no objections to resolving it with this text
RESOLUTION: accept text in the bug report
thanks asir :)
http://www.w3.org/Bugs/Public/show_bug.cgi?id=3712
<asir> related editorial action for 3711 - http://www.w3.org/2005/06/tracker/wspolicyeds/actions/40
<PaulC> http://www.w3.org/Bugs/Public/show_bug.cgi?id=3712
<vladB> http://lists.w3.org/Archives/Public/public-ws-policy/2006Sep/0083.html
fred: sounds like it trats policyReference as a standalone definition
paul: that is the intent
asir: policyReference is a global element in schema
framework doesn't provide semantics
asir: how it applies to others such as wsdl 2.0, uddi, etc
paul: anybody that wants to use policyReference has to define the semantics where it's used
vladB: this is an artificial division between the two specs. if I receive a wsdl with this specification, and i don't see it here
paul: vlad is saying when I see a policyReference mentioned, it's not clear which spec it's referring to
asir: when it's near an attachment point, attachment spec applies. when it's elsewhere, it's framework
vladB: maybe we need two different elements
maryann: could we define it more clearly here?. the words you said are not in this spec
fred: it seems that what's being said is that policyReference is a first class. in section 3 of framework we could have a section between 3.2 and 3.3 that talks about policyReference. explains how it's used. other documents would refer to it
asir: section 3 is about data model. no reference at this level
fred: maybe should be between sec 3 and 4. where would i put it in the table of contents?
paul: at the heart of the proposal and request is to pull the text and make it standalone. so both specs can refer to it. i agree it belongs in section 4, not 3. need a proposal to take it out, and add a statement that mentions that the semantics are defined elsewhere, and refer to it
ACTION: asir to provide a proposal for 3712 [recorded in http://www.w3.org/2006/09/14-ws-policy-minutes.html#action06]
<trackbot> Created ACTION-114 - Provide a proposal for 3712 [on Asir Vedamuthu - due 2006-09-21].
Duplicates (committed twice?)
RESOLUTION: 3722 is duplicate of 3719
paul: he's asking for more levels of indirection. e.g. policy reference to uddi entry which points to another policy
does the current spec restrict you?
the current specs combined (framework and attachment) provide for that requirement
<Ashok> ACTION: danroth to respond back explaining the description we just had, and asking whether that satisfies the requirement [recorded in http://www.w3.org/2006/09/14-ws-policy-minutes.html#action09]